Montréal Mars 2008

© 2008 Ari Van Assche, Galina A. Schwartz. Tous droits réservés. All rights reserved. Reproduction partielle permise avec citation du document source, incluant la notice ©. Short sections may be quoted without explicit permission, if full credit, including © notice, is given to the source.

Série Scientifique Scientific Series

2008s-07

Institutions and Multinational Ownership Strategy

Ari Van Assche, Galina A. Schwartz

CIRANO Le CIRANO est un organisme sans but lucratif constitué en vertu de la Loi des compagnies du Québec. Le financement de son infrastructure et de ses activités de recherche provient des cotisations de ses organisations-membres, d’une subvention d’infrastructure du Ministère du Développement économique et régional et de la Recherche, de même que des subventions et mandats obtenus par ses équipes de recherche.

CIRANO is a private non-profit organization incorporated under the Québec Companies Act. Its infrastructure and research activities are funded through fees paid by member organizations, an infrastructure grant from the Ministère du Développement économique et régional et de la Recherche, and grants and research mandates obtained by its research teams. Les partenaires du CIRANO Partenaire majeur Ministère du Développement économique, de l’Innovation et de l’Exportation Partenaires corporatifs Alcan inc. Banque de développement du Canada Banque du Canada Banque Laurentienne du Canada Banque Nationale du Canada Banque Royale du Canada Banque Scotia Bell Canada BMO Groupe financier Bourse de Montréal Caisse de dépôt et placement du Québec DMR Conseil Fédération des caisses Desjardins du Québec Gaz de France Gaz Métro Hydro-Québec Industrie Canada Investissements PSP Ministère des Finances du Québec Raymond Chabot Grant Thornton State Street Global Advisors Transat A.T. Ville de Montréal Partenaires universitaires École Polytechnique de Montréal HEC Montréal McGill University Université Concordia Université de Montréal Université de Sherbrooke Université du Québec Université du Québec à Montréal Université Laval Le CIRANO collabore avec de nombreux centres et chaires de recherche universitaires dont on peut consulter la liste sur son site web.

ISSN 1198-8177

Les cahiers de la série scientifique (CS) visent à rendre accessibles des résultats de recherche effectuée au CIRANO afin de susciter échanges et commentaires. Ces cahiers sont écrits dans le style des publications scientifiques. Les idées et les opinions émises sont sous l’unique responsabilité des auteurs et ne représentent pas nécessairement les positions du CIRANO ou de ses partenaires. This paper presents research carried out at CIRANO and aims at encouraging discussion and comment. The observations and viewpoints expressed are the sole responsibility of the authors. They do not necessarily represent positions of CIRANO or its partners.

Partenaire financier

Institutions and Multinational Ownership Strategy*

Ari Van Assche†, Galina A. Schwartz ‡

Résumé / Abstract Ce papier examine l'impact des institutions sur la stratégie de propriété d’une entreprise multinationale. Nous développons un modèle de coentreprise internationale dans lequel une entreprise étrangère et son partenaire local peuvent ex post entreprendre des actions coûteuses pour augmenter leur part de revenus indiquée dans le contrat de coentreprise. Le modèle analyse les impacts de deux caractéristiques institutionnelles sur la structure de propriété optimale : le renforcement du contrat et le copinage. Nous introduisons le modèle de coentreprise dans un modèle d'équilibre général pour analyser l'impact des institutions sur le mode d’entrée des entreprises multinationales.

Mots clés : coentreprise internationale, renforcement du contrat, copinage.

This paper examines the impact of institutions on a multinational firm’s ownership strategy. We develop an international joint venture (IJV) model in which a multinational firm and its local partner both can undertake costly ex post actions to increase their revenue share specified by the ex ante IJV contract. The model captures the effects of two institutional features on the optimal IJV ownership structure: contract enforceability and cronyism. We introduce the IJV model into an industry equilibrium framework to analyze the impact of institutions on a multinational firm’s choice between forming an IJV or setting up a wholly-owned subsidiary.

Keywords: liability of foreignness, international joint venture, contract enforceability, cronyism. Codes JEL : F23, F12.

* We thank Phelim Boyle, Nathan Jensen, Justin Leroux, Suzanne Rivard, Nicolas Sahuguet and seminar participants at the Université de Québec à Montréal, Carleton University and the Katholieke Universiteit Leuven for helpful comments and suggestions. Ari Van Assche thanks the fonds québécois de la recherche sur la société et la culture (FQRSC) for financial support. † HEC Montréal, Department of International Business, 3000, Chemin de la Côte-Sainte-Catherine, Montréal (Québec), Canada, H3T2A7. Phone: (514)340-6043. Fax: (514)340-6987. E-mail: ari.van-assche@hec.ca. ‡ University of California, Berkeley.

1 Introduction

A long-standing stylized fact in international economics is that firms face aliability of foreignness (LOF) when doing business abroad (Hymer, 1976).1

Foreign firms have less information about the host country than local firms.They may also receive inferior treatment from the host country governmentdue to their relative lack of connections with local officials. LOF has becomea key building block in theories of the multinational firm, which motivatesthe paradigm that multinational firms need firm-specific advantages to com-pete successfully against local firms (Buckley and Casson, 1976; Caves, 1982;Dunning, 1977; Hennart, 1982).

Despite the importance of LOF in theories of the multinational firm,formal studies of its effects on a multinational firm’s entry strategy arescant.2 The effects are not straightforward. Forming an international jointventure (IJV) with a local firm is generally considered to mitigate LOF byallowing a multinational firm to tap into its local partner’s resources andconnections (Hennart, 1988, 1991; Inkpen and Beamish, 1997; Luo, 2002).The multinational firm can for example use its local partner’s relations withgovernment officials to obtain the required business licenses to set up theIJV. Less considered, however, is the fact that once a multinational firmis locked into an IJV, its local partner’s superior market knowledge andcontacts remain. In weak institutional environments, a local partner thenmay have the possibility to use this ex post advantage to get extra privatebenefits from the IJV at the detriment of the multinational firm (Henisz,2000; Perkins et al., 2007).

The following example by Desai and Moel (2008) demonstrates the effectsof such an ex post LOF. In the early 1990s, the U.S.-invested enterprise CMEformed an IJV with the Czech company CET-21 to enter the Czech mediamarket. This joint venture was considered beneficial since it should reduceCME’s ex ante LOF. Indeed, CME counted on the strong market knowledgeof CET-21’s owner Vladimir Zelezny to ensure locally appropriate televisionprogramming.3 Moreover, by forming an IJV, CME wanted to circumventthe political opposition against giving television licenses to foreign investorsby assigning the ownership of the license to CET-21. The partners’ IJVcontract then specified that CME had exclusive user rights over the license.

1In recent work, however, Huang (2005) has found that foreign firms often enjoy regu-latory advantages over weakly connected local firms.

2See Zaheer (1995, 2002) and Eden and Miller (2004) for recent surveys of the inter-national business literature on this topic.

3An interesting anecdote: Zelezny means “man of iron” in Russian.

2

The IJV’s television station launch was highly successful. In its first year ofoperation, it gained a 70 percent share of the Czech television audience. Inthe second half of the 1990s, however, the partnership turned sour. At thattime, Zelezny successfully used his government contacts to obtain a rulingthat CME could no longer have exclusive use of the license. This decisionallowed CET-21 to expropriate CME’s rents from the joint venture.

The example demonstrates that well-connected local firms in weak insti-tutional environments may have the ability to manipulate local officials toexpropriate rents from their multinational partners (Desai and Moel, 2008).This ability affects a multinational firm’s ownership strategy in two ways.First, it reduces the attractiveness of entering a foreign market through jointventures with local firms. Second, it affects the ex post distribution of rentsbetween IJV partners, thus distorting their incentives to contribute specificinvestments. Therefore, it also affects the optimal IJV ownership structure.

Local firms’ ability to ex post expropriate rents depends on the hostcountry’s institutional environment. Specifically, two institutional featuresneed to coexist. On one hand, cronyism needs to exist so that local officialsgive preference to well-connected local firms. On the other hand, IJV con-tracts need to be imperfectly enforceable so that the local firm can ex posttake actions to take advantage of cronyism.

This paper examines the impact of these institutional features on a multi-national firm’s ownership strategy. In Section 2, we set up an IJV model inwhich a multinational firm first (ex ante) signs a revenue-sharing contractwith a local firm. Then each party provides specialized inputs to the IJV tocreate joint revenue. Finally (ex post), parties can conduct costly actions toincrease their revenue share. Such rent-seeking actions include governmentlobbying, gifts and bribes.

Our model captures the institutional features of contract enforceabil-ity and cronyism by modelling IJV partners’ “bribing” environments. Thestricter a country’s legal system is in enforcing contracts, the lower the in-centives for parties to violate the stipulations of the original contract. In ourmodel, we thus represent contract enforceability with a parameter that cor-responds to parties’ costs of ex post taking rent-seeking actions. To modelcronyism, we let the cost of these ex post actions be lower for local firmsthan for multinational firms. An increase in disparity between parties’ costsof ex post actions then implies more pervasive cronyism.

The results of Section 2 are as follows. Both contract enforceabilityand cronyism affect a multinational firm’s ownership strategy. A reductionin contract enforceability negatively affects IJV performance by (i) creatingwasteful ex post expenses and (ii) distorting partners’ investment incentives.

3

In distorting incentives, it disproportionately weakens those of the minoritypartner (i.e., the partner with a smaller ownership share). We find thatit is thus optimal to deal with lower contract enforceability by allocatinga larger ownership share to the minority partner. Cronyism also distortspartners’ investment incentives by strengthening those of the local firm whileweakening those of the multinational firm. We find that it is thereforeoptimal to deal with cronyism by allocating a higher ownership share to themultinational firm. Interestingly, we find that cronyism does not necessarilyreduce IJV performance. Rather, it lowers the feasibility of an IJV. Insum, IJV partners need to take into account both institutional features todetermine their optimal IJV ownership structure.

In Section 3, we introduce our IJV model into an industry equilibriumframework to analyze the impact of the institutional features on a multina-tional firm’s entry mode. In this framework, heterogeneous multinationalfirms choose from two modes: (i) they can form an IJV or (ii) they can setup a wholly-owned subsidiary. There is an intrinsic trade-off between thesemodes. If a multinational firm forms an IJV, it gains a lower fixed cost ofentry since it can tap into the local partner’s market-specific knowledge andcontacts. But it faces higher transaction costs due to imperfect contractenforceability and cronyism. Inversely, setting up a subsidiary permits it togain lower transaction costs at a higher fixed entry cost. We find that formore productive multinational firms, setting up a wholly-owned subsidiaryis optimal, and forming the IJV is optimal for less productive firms.

Our paper is related to recent theoretical studies on the optimal revenue-sharing contract in joint ventures. Bai et al. (2004) and Wang and Zhu(2005) set up joint venture models with incomplete contracting in whichownership and control are separately contractible. In these papers, owner-ship determines revenue-sharing, while control determines parties’ ability toex post take actions to acquire private benefits at a loss to the joint ven-ture. Their setups allow them to simultaneously study the optimal structureof revenue sharing and control allocation in a joint venture. Our paper issimilar to theirs in that we both study optimal revenue-sharing contracts injoint ventures. It differs, however, in that we do not model the allocationof control rights, but rather focus on the role of the institutional featureson ex post actions. That is, ex post actions in our model are not driven bythe allocation of control rights, but rather by the “bribing” environments inwhich IJV partners operates.

Our paper also builds onto the recent literature on contract enforcement

4

and multinational firms’ organizational decisions.4 This literature incorpo-rates elements of incomplete contracts theory into general-equilibrium trademodels to analyze the determinants of multinational firms’ decisions to off-shore and/or outsource their production activities. Our benchmark modelclosely resembles that of Antras and Helpman (2004). By introducing costlyex post actions into it, we contribute to this literature new analytical un-derpinnings to study the impact of the institutional features of contractenforceability and cronyism on a multinational firm’s organization. In thispaper, we apply our setup to study the multinational firm’s optimal owner-ship strategy in a foreign market.

The rest of the paper is organized as follows. Section 2 sets up the IJVmodel. Section 3 incorporates the IJV model into an industry-equilibriumframework. Section 4 concludes.

2 IJV Model

Consider a multinational firm M that sets up an IJV with a local firm D.Let the IJV face an iso-elastic inverse demand function for its output y:

p = λ1−αyα−1, (1)

where p denotes price, the constant α ∈ [0, 1/2] determines the elasticity ofdemand and the constant λ > 0 reflects the level of demand.5

Production of final good y requires two complementary inputs: foreigninputs m produced by the multinational firm at unit cost ρ and local inputsd produced by the local firm at unit cost ω. We assume that the inputsare relationship-specific, that is, they have zero outside value if the IJVrelationship breaks down. The IJV produces final good y from inputs mand d via a Cobb-Douglas production function:6

y = θ

(m

η

)η (d

1− η

)1−η

, (2)

where the constant θ reflects the IJV’s productivity and η ∈ [0, 1] is aparameter characterizing the intensity of foreign inputs in IJV production.

4This literature includes studies by Grossman and Helpman (2002, 2003, 2005); Antras(2003, 2005) and Antras and Helpman (2004, 2008). See Spencer (2005) and Helpman(2006) for recent surveys of this literature.

5We assume that α ∈ [0, 1/2] to guarantee the uniqueness of an equilibrium in section2.3. See Appendix for more details.

6We have purposefully adopted many of the functional forms and notation from Antrasand Helpman (2004) to ease comparison with the results of their paper.

5

By combining (1) and (2), the IJV’s revenue can be expressed as

R(m, d) = λ1−αθα

(m

η

)αη (d

1− η

)α(1−η)

. (3)

We assume that revenue R is verifiable by an outside party, but that thequality of inputs is unverifiable. To induce the parties to provide the re-quired specific inputs, they can sign a linear revenue-sharing contract thatlinks their incomes to the verifiable joint revenue.7 Specifically, the multina-tional firm can propose a take-it-or-leave-it IJV contract to a local firm thatspecifies (i) the multinational firm’s ownership share s ∈ [0, 1] that gives itright to revenue share s and (ii) a fixed lump-sum payment t between thetwo parties that takes place ex ante, i.e. prior to the production of inputs.

The IJV operates in an institutional environment where parties can expost take costly actions to adjust the ex ante determined allocation of rev-enue. After parties have produced their inputs and revenue is realized, eachparty i ∈ M,D can spend private resources or “bribes” to increase itsrevenue share by ri.8 Let v and 1− v denote the ex post revenue share forthe multinational and local firm respectively. Parties ex post choose rm andrd so that the multinational firm’s ex post revenue share

v = s+ rm − rd. (4)

We assume that the ex post actions are costly. To improve its ex postrevenue share with ri, a party incurs the bribing expense B(ri), which isincreasing and concave in adjustment ri:

B(ri) =

e

riγi if ri > 00 ri = 0.

(5)

In the “bribe function” B, the parameter γi reflects the bribing environmentin which party i operates.9 The higher γi, the lower the bribe B that party ipays to increase its revenue share by ri. We infer from this that an increase

7Such linear contracts are widespread in business practices due to their their simplicityand transparency. They are also widely adopted by the literature on joint ventures, seeBai et al. (2004) and Wang and Zhu (2005).

8We for simplicity call these rent-seeking actions “bribes”. More accurately, theseactions encompass all legal and illegal ex post transactional activities including legal ex-penses, government lobbying and gifts.

9Our bribe function implies that a party has to pay a fixed cost of 1 to bribe. Thisassumption can easily be generalized to account for a fixed bribing cost of any size.

6

in party i’s bribing parameter γi strengthens its incentives to undertake expost actions.

The IJV parties’ bribing parameters γi capture two features of the hostcountry’s institutions: (i) contract enforceability and (ii) cronyism. First, weuse the average of both parties’ bribing parameters as a measure of contractenforceability. Let

γ =γm + γd

2. (6)

An increase in γ then reflects lower contract enforceability. The intuitionbehind this is the following. If γ is high (low), it is cheap (expensive) fora party to ex post violate the terms of the ex ante contract. As a result,parties have weak (strong) incentives to adhere to the ex ante contract. High(low) γ thus corresponds to low (high) contract enforceability.

Second, we can use the difference between parties’ bribing parameters γi

to characterize cronyism. Let

δ =γd − γm

2. (7)

If δ > 0, the cost of ex post actions is lower for a local firm than for amultinational firm (γd > γm). This corresponds to an environment in whichofficials give special treatment to well-connected local firms. We use δ tomeasure cronyism, with a higher δ reflecting more pervasive cronyism.

Note that combining (6) and (7) yields

γd = γ + δ and γm = γ − δ, (8)

where γ ≥ δ ≥ 0. We infer from the latter condition that cronyism isdependent on contract enforceability. Specifically, cronyism (δ > 0) canonly be present if contracts are imperfectly enforceable (γ > 0). Below, weuse (8) to insert the institutional parameters γ and δ into the model.

The IJV model can be summarized by the following sequence of events.In period 1, the multinational firm offers a take-it-or-leave-it IJV contract(s, t) to a local firm that specifies the multinational firm’s ownership shares and a lump-sum transfer t. In period 2, both parties produce their inputsm and d. In period 3, both parties can adjust the ex ante contract throughcostly actions rm and rd. Below, we solve for the optimal IJV contractthrough backward induction.

The remainder of this section is divided into three parts. In subsection2.1, we analyze the benchmark case of perfect contract enforcement whereit is too costly for parties to conduct ex post actions. In subsection 2.2, we

7

study the case of imperfect contract enforcement without cronyism to ana-lyze the impact of contract enforceability on the optimal IJV contract andperformance. In subsection 2.3, we analyze the case of imperfect contractenforcement with cronyism to assess the impact of cronyism on the optimalIJV contract and performance.

2.1 Perfect Contract Enforcement

Consider the benchmark case of perfect contract enforcement where γi = 0.For both parties, it is then infinitely costly to ex post conduct rent-seekingactions. As a result, it is optimal for each party i in period 3 to choose

ri = 0,

where the superscriptˆdenotes optimum under perfect contract enforcement.Both parties’ ex post revenue shares then coincide with their ownershipshare:

v = s.

Next, consider the parties’ choices in period 2. Here they choose profit-maximizing inputs m and d for a given contract (s, t):

maxm

πm = sR(m, d)− ρm+ t,

maxdπd = (1− s)R(m, d)− ωd− t,

where πi denotes party i’s profits. By solving these problems and insertingthem into (3), we can express revenue R as a function of the multinationalfirm’s ownership share s:

R(s) = λ

[αθ

(s

ρ

)η (1− s

ω

)1−η] α

1−α

.

In period 1, the multinational firm offers a take-it-or-leave-it contract (s, t)to its local partner that guarantees the local partner’s participation in theIJV. It thus solves

maxs,t

πm = sR(s)− ρm(s) + t (9)

subject to

πd = (1− s)R(s)− ωd(s)− t ≥ 0.

8

By solving for the optimal contract (s, t), we derive an expression for amultinational firm’s optimal ownership share:

s =

1/2 if η = 1/2η−Ψ2η−1 if η 6= 1/2 (10)

whereΨ = αη(1− η) +

√η(1− η)(1− αη)(1− α+ αη). (11)

It is straightforward to show that Ψ ∈ [0, 1/2] and that Ψ < η if η < 1/2,while Ψ > η if η > 1/2. As a result, s ∈ [0, 1].

[Figure 1 about here]

In Figure 1, we use (10) and (11) to plot as a solid curve the relation betweenownership share s and the intensity of foreign inputs η. Ownership share sapproaches zero when η approaches 0, it approaches 1 when η approaches1, and it rises in between. Moreover, s is concave for η < 1/2 and convexfor η > 1/2. These properties reflect that parties cannot contract uponthe purchase of specialized inputs for a certain price. As a result, neitherparty appropriates the full marginal return to its investments in the supplyof inputs, thus leading to underinvestment. Each party’s incentives arethen determined by its ownership share s. Ex ante efficiency thus requiresgiving a larger ownership share to the party undertaking the relatively moreimportant investment into the IJV. As a result, the higher the intensity offoreign inputs (the larger η is), the higher the multinational firm’s ownershipshare s is. This leads to Proposition 1:

Proposition 1 Ceteris paribus, under perfect contract enforcement, themultinational firm’s ownership share in an IJV rises with foreign input in-tensity.

Next, we can derive πm by inserting the optimal contract (s, t) into (9). Thisgives us

πm = λΩ, (12)

whereΩ =

(1− α(1−Ψ))(2η−1αθ

(ρ

η−Ψ

)η (ω

η+Ψ−1

)1−η) α

1−α

. (13)

9

Our benchmark model with perfectly enforceable contracts replicates a keyresult of Antras and Helpman (2004).10 Next, we will use this benchmark toanalyze the impact of the institutional features contract enforceability andcronyism on a multinational firm’s ownership strategy.

2.2 Contract Enforceability

Consider the case of imperfect contract enforcement (γ > 0) without crony-ism (δ = 0). In this case, it is optimal for both parties to ex post engage inrent-seeking actions.

First, we derive parties’ optimal ex post actions for a given (s, t,m, d).Each party chooses the optimal amount ri that solves:

maxrm

πm = (s− rd + rm)R−B(rm)− ρm+ t (14)

maxrd

πd = (1− s+ rd − rm)R−B(rd)− ωd− t, (15)

where the bribe function B(ri) is given by (5) and R is short for R =R(s, t,m, d). The solutions of these problems yield

r∗i = γ ln (γR), (16)

where superscript * reflects optimum under imperfect contract enforcement.Inserting (16) into (5) then determines parties’ bribes

B(r∗i ) = γR.

This suggests that both parties spend fraction γ of revenue R on bribes toincrease their revenue share by r∗i .

11 Since r∗i is identical for both parties,their effects on v cancel each other out. Thus, each party’s ex post revenueshare is identical to its ownership share:

v∗ = s.

In period 2, parties choose inputs m and d for a given (s, t):

maxm

πm = (s− γ)R(m, d)− ρm+ t

10Our equations (10) (11), (12) and (13) are identical to Antras and Helpman’s (2004)equations (6), (7) and (10).

11Note that the IJV becomes unfeasible if γ ≥ 12. In that case, for at least one of both

parties, the share of revenue that it needs to pay as a bribe γ exceeds its ownership share.In our discussion below, we assume that γ ≤ 1

2.

10

maxdπd = (1− s− γ)R(m, d)− ωd− t.

By solving these problems and inserting them into (3), we can derive revenueas a function of s:

R∗(s) = λ

[αθ

(s− γ

ρ

)η (1− s− γ

ω

)1−η] α

1−α

. (17)

It is straightforward to show from (17) that R∗ decreases with γ. The reasonis that imperfect contract enforceability negatively affects both parties’ in-centives to produce inputs. When choosing their optimal amount of inputs,each party takes into account that, ceteris paribus, producing more inputsincreases revenue and thus raises bribe γR. This acts as a disincentive forparties to produce inputs.

The effect of imperfect contract enforceability on parties’ incentives isnot identical. While an increase in revenue induces both parties to increasetheir bribes equally, the ex post distribution of the extra revenue is notnecessarily even. Specifically, the multinational firm ex post receives share sof the extra revenue, while the local firm receives share 1− s. We infer fromthis that the incentives of the minority partner in an IJV (i.e., the partnerwith a smaller ownership share) are affected more negatively by imperfectcontract enforceability than those of the majority partner. Below, we showthat this distortion of incentives affects the optimal ownership structure.

In period 1, the multinational firm offers a take-it-or-leave-it contract(s, t) to the local firm that solves the problem

maxs,t

πm = (s− γ)R(s)− ρm(s) + t (18)

subject to

πd = (1− s− γ)R(s)− ωd(s)− t ≥ 0.

By combining the solution of this problem with (10) and (11), we can relatethe multinational firm’s ownership share s∗ with its ownership share underperfect contract enforcement s:

s∗ = s− γ1− 2Ψ2η − 1

. (19)

In Figure 1, we use (19) to plot the relation between ownership share s∗ andthe intensity of foreign inputs η as a dashed line. Similar to the benchmark

11

case, s∗ increases with η.12 Contract enforceability affects the intersect andslope of the curve however. First, to ensure that both parties receive asufficiently large share of revenue to pay the bribe γR, s∗ approaches γwhen η approaches 0 and s∗ approaches 1−γ when η approaches 1. Second,a comparison of (10) with (19) shows that the slope of s∗ is less steep thanthe slope of s and that the steepness of s∗ is positively related to contractenforceability 1/γ. This allows us to infer the following Proposition:

Proposition 2 When the multinational firm contributes relatively more tothe IJV (η > 1/2), its ownership share decreases with γ. Otherwise (η <1/2), its ownership share increases with γ.

The intuition behind Proposition 2 is the following. Lower contract enforce-ability disproportionately reduces the incentives of the minority partner inthe IJV. Since the multinational firm is the minority partner in a local-inputintensive IJV (η < 1/2), it is optimal to allocate a larger ownership share tothe multinational firm to improve its incentives. In a foreign-input intensiveIJV (η > 1/2), the local firm is the minority partner. In this case, it isoptimal to allocate a larger ownership share to the local partner to provideit with more powerful incentives.

Next, we analyze the effect of contract enforceability on the efficiency ofan IJV. By inserting the optimal contract (s∗, t∗) into (18) and combiningit with (12) and (13), we obtain:

π∗m = λΩ (1− 2γ)1

1−α = πm (1− 2γ)1

1−α . (20)

Equation (20) suggests that a multinational firm receives share (1− 2γ)1

1−α

of its profits under perfect contract enforcement. This share decreases inγ and is not affected by foreign input intensity η. From our discussion, anincrease in γ reduces the efficiency of an IJV through two channels. First,it reduces parties’ incentives to produce inputs, thus reducing joint revenue.Second, it increases the share of revenue that both parties waste on their expost actions.

Interestingly, foreign input specificity η does not affect the efficiency lossthat an IJV faces. This observation together with Proposition 2 suggeststhat an IJV in an industry with a low or high foreign input intensity η canbe equally efficient as an IJV with an intermediate value of η as long as theownership structure is sufficiently adjusted to provide the minority partnerwith higher-powered incentives.

12Since Ψ ≤ 12

and 12≥ γ ≥ 1

R, s > s∗ ≥ 0 in local-input intensive industries (η < 1/2)

and 1 ≥ s∗ > s in foreign-input intensive industries (η > 1/2).

12

2.3 Cronyism

Finally, we consider the case of cronyism. Let δ > 0 so that the cost ofex post adjustment is higher for the multinational firm than for its localpartner (γd > γm). In period 3, both parties choose optimal ri for a given(s, t,m, d). Solving (14) and (15) yields:

r~m = (γ − δ) ln ((γ − δ)R) (21)r~d = (γ + δ) ln ((γ + δ)R) (22)

where subscript ~ reflects optimum in an environment with cronyism. Com-bining (21) and (22) with (4) gives

v~ = s−∆ (23)

where∆ = r~

d − r~m ≥ 0 and ∆ = ∆(R). (24)

We infer from (23) and (24) that cronyism allows the local partner to ex postgrab revenue share ∆ from the multinational firm. It is straightforward toderive that ∆ increases with R. As we demonstrate below, this implies thatcronyism affects parties’ incentives to contribute inputs.

In period 2, parties choose the inputs m and d for a given (s, t). We useequations (21)-(24) to rewrite parties’ maximization problems as:

maxm

πm = (s−∆(R(m, d))− γ + δ)R(m, d)− ρm+ t (25)

maxdπd = (1− s+ ∆(R(m, d))− γ − δ)R(m, d)− ωd− t. (26)

From (23) - (26), cronyism affects parties’ incentives to produce inputs, al-beit differently. It reduces the multinational firm’s incentives to produceinputs since the multinational firm considers that, ceteris paribus, produc-ing more m increases R and thus increases the share ∆ that the local partnergrabs. Inversely, cronyism increases the local firm’s incentives since produc-ing more inputs d increases R and thus raises the share ∆ that it grabs fromthe multinational firm.

In period 1, the multinational firm offers a take-it-or-leave-it contract(s, t) to its local partner. In the appendix, we derive that it is optimal for amultinational firm to choose ownership share s~ that allows it to obtain expost revenue share v~ :

v~ = s− γ1− 2ψ2η − 1

+ δ. (27)

13

We also demonstrate that there is a unique s~ that corresponds to v~.If we compare (19) with (27), v~ consists of three terms. The first term

is the multinational firm’s optimal ownership share under perfect contractenforcement s; the second term reflects the effect of imperfect contract en-forceability; and the third term reflects the effect of cronyism. As is clearfrom (27) and as is depicted with a dashed and pointed line in Figure 1,a multinational firm thus should respond to an environment with cronyism(δ > 0) by ensuring that its ex post revenue share increases with δ. As wehave explained above, this is because cronyism reduces the multinational’sincentives to produce inputs, while it increases the local firm’s incentivesto produce inputs. Ensuring that the multinational firm’s ex post revenueshare increases with δ thus allows the IJV to increase the relative power ofthe multinational firm’s incentives.

To obtain v~, a multinational firm needs to choose ownership shares~. By combining (23) with (24) and (27), it is straightforward to showthat, ceteris paribus, the multinational firm’s ownership share is larger inan environment with cronyism than in an environment without cronyism:

s~ ≥ s∗ = s− γ1− 2ψ2η − 1

. (28)

We state this in the following Proposition:

Proposition 3 Ceteris paribus, a multinational firm’s ownership share inan IJV is higher in an environment with cronyism (δ > 0) than in anenvironment without cronyism (δ = 0).

Note that the multinational firm cannot always obtain v~ through its choiceof s~. Specifically, if v~ + ∆ > 1, it is required to choose s~ > 1, whichis not possible. In this case, it is profit-maximizing to choose s~ = 1. Butthis exclusive ownership goes against the notion of a joint venture. We inferthat cronyism negatively affects the feasibility of forming an IJV. Two typesof IJVs are particularly prone to become unfeasible in an environment withcronyism. First, an IJV with a higher foreign-input intensity η is more likelyto become unfeasible since this type of joint venture requires a higher v~,thus increasing the likelihood that s~ = 1. Second, an IJV with higherproductivity θ is more likely to become unfeasible since this type of jointventure has a higher R and therefore a higher ∆. This also increases thelikelihood that s~ = 1. We state this in the following Proposition:

Proposition 4 Ceteris paribus, cronyism reduces the feasibility of an IJV.Feasibility is especially reduced when η and θ are high.

14

Finally, we in the appendix derive the multinational firm’s profits undercronyism:

π~m = π∗m = πm (1− 2γ)

11−α . (29)

Interestingly, equation (29) suggests that cronyism does not entail an extraefficiency loss. That is, for s~ < 1, the multinational firm’s profits areidentical to its profits with δ = 0 (equation (20)). Still, from Proposition 4,we expect fewer IJVs to be formed when cronyism is present since it reducesthe feasibility of an IJV.

3 Entry mode

In this Section, we introduce the IJV model into an industry equilibriumframework similar to Antras and Helpman (2004) to analyze the impactof the institutional features on a multinational firm’s optimal entry mode.Consider an industry-equilibrium model where a continuum of multinationalfirms decide on their entry mode into a foreign market. We assume that themultinational firms need to be physically located in the host country to selltheir products.13 Local consumers are assumed to spend a constant shareof their income on the multinational firm’s products. They have Dixit-Stiglitz preferences that gives rise to the inverse demand function (1), whereλ depicts the aggregate consumption index. Since there is a continuum offirms, each multinational firm takes λ as given.

For the production of a final good variety, two parties are required: amultinational firm that produces inputs m and a local entity that producesinputs d. Parties face a perfectly elastic supply of the unique factor ofproduction, labor. Wages are fixed so that the cost of producing a foreigninput is parameter ρ and the cost of producing a local input is parameter ω.Final goods are produced using the Cobb-Douglas production function (2).

Similar to Melitz (2003), multinational firms differ in their productivity.To learn its productivity, a multinational firm incurs an irreversible fixed costof entry equal to Fe. Upon paying this fixed cost, it learns its productivitylevel θ, which is randomly drawn from a known cumulative distributionfunction G(θ). After observing its productivity level, it then decides whetherto enter the host country market. Multinational firms have two entry modeoptions. First, to form an IJV with a local firm (denoted by superscript J).Second, to set up a wholly-owned subsidiary (denoted by superscript V ).

13This assumption rules out market entry through exports from a multinational firm’shome country.

15

We assume that under both entry modes, a multinational firm signs a linearrevenue-sharing contract (s, t) with a local entity.

There is a key trade-off between both entry modes. On one hand, formingan IJV entails a lower fixed cost than setting up a subsidiary since it providesthe multinational firm with access to local firms’ market specific knowledgeand contacts: κV > κJ . On the other hand, setting up a subsidiary implieslower transaction costs since, unlike for an IJV, contracts are consideredperfectly enforceable under this entry mode.

The industry-equilibrium model can be summarized by the followingsequences of moves: in period 0, each multinational firm decides whetherit enters the host country. If it enters, it incurs a fixed cost Fe to have itsproductivity level θ realized. In period 1, the multinational firm decides ifit wants to produce output or remain idle. If it decides to produce output,it chooses to enter the market by forming an IJV with a local partner orby setting up its own subsidiary. In period 2, the multinational firm signs arevenue-sharing contract (s, t) with its local partner or subsidiary. In period3, both parties produce their inputs. In period 4, IJV partners can takecostly ex post actions to improve their revenue share. The final goods arethen produced and sold.

We can solve the model through backward induction. For periods 2-4,we can rely on our derivations in Section 2. Specifically, we can use (20)to infer that if a multinational firm signs an imperfectly enforceable IJVcontract with a local firm, it earns the following profits:

πJm = λΩ (1− 2γ)

11−α − κJ . (30)

Similarly, we can use (12) to infer that if a multinational firm signs a per-fectly enforceable contract with its subsidiary, it earns profits:

πVm = λΩ− κV . (31)

In period 1, each multinational firm considers profits functions (30) and(31) to decide its optimal entry mode. In Figure 2, we plot both profitfunctions to graphically analyze the role of productivity on a multinationalfirm’s optimal entry mode. We depict θ

α1−α on the horizontal axis and the

multinational firm’s profit on the vertical axis. In addition, we define θ1 asthe productivity level where πJ

m = πVm.

[Figure 2 about here]

16

Figure 2 demonstrates that less productive firms with θ ≤ θ1 choose to forman IJV with a local partner, while more productive firms with θ > θ1 chooseto set up a wholly-owned subsidiary. This suggests that less productivemultinational firms choose to focus on the mitigation of ex ante LOF byforming an IJV, while more productive multinational firms focus on themitigation of transaction costs by setting up a wholly-owned subsidiary. Westate this in the following Proposition:

Proposition 5 Ceteris paribus, forming an IJV is optimal for low produc-tivity firms with θ ≤ θ1; setting up a wholly-owned subsidiary is optimal forhigh productivity firms with θ > θ1.

4 Conclusion

The question at the heart of this paper is how multinational firms shouldadapt their ownership strategy when facing the institutional features of con-tract enforceability and cronyism. To address this, we have set up an IJVmodel where both the multinational firm and its local partner can ex posttake costly actions to increase their revenue share. We have shown that theIJV’s optimal ownership structure in such a model is determined by bothinstitutional features. Specifically, lower contract enforceability (higher γ)disproportionately weakens the incentives of the minority partner in an IJV.As a result, it is optimal for a multinational firm to allocate a larger owner-ship share to the minority partner in the IJV. Cronyism (higher δ) weakensthe incentives of the multinational firm to contribute inputs to the IJV whilestrengthening the incentives of the local firm. The multinational firm thusshould receive a larger ownership share in the IJV.

Next, we have introduced the IJV model into an industry-equilibriumsetup to study the effect of the institutional features on a multinationalfirm’s entry mode. We have demonstrated that, in an environment with lowcontract enforceability and cronyism, more productive multinational firmswill set up their own subsidiary, but that less productive multinationalswill form an IJV. We have also shown that lower contract enforceability andhigher cronyism both increase the prevalence of wholly-owned subsidiaries inan industry. Our analysis identifies testable hypotheses relating institutionalfeatures with a multinational firm’s optimal ownership strategy. This opensthe door for empirical studies.

17

5 Appendix

In this appendix, we solve the IJV model with cronyism (section 2.3) throughbackward induction.

Period 3. For a given (s, t,m, d), parties solve

maxrm

πm = (s− rd + rm)R− ermγ−δ − ρm+ t (A-1)

maxrd

πd = (1− s+ rd − rm)R− erd

γ+δ − ω.d− t (A-2)

The solution of these problems gives:

r~m = (γ − δ) (lnR+ ln (γ − δ)) (A-3)r~d = (γ + δ) (lnR+ ln (γ + δ)) . (A-4)

To simplify notation below, let

∆(R) = r~d − r~

m = 2δ lnR+ (γ + δ) ln (γ + δ)− (γ − δ) ln (γ − δ). (A-5)

Then, the multinational firm’s ex post revenue share is

v = s−∆(R). (A-6)

Period 2. For a given (s, t), parties choose inputs m and d. By inserting(A-3) and (A-4) into (A-1) and (A-2), their maximization problems can bereduced to:

maxm

πm = (s−∆(R)− γ + δ)R− ρm+ t (A-7)

maxdπd = (1− s+ ∆(R)− γ − δ)R− ωd− t, (A-8)

where R is a shortcut for R(m, d), which is given by (3).Next, we will demonstrate that when α ∈ (0, 1/2) for any fixed s, there

exists a unique optimal m~ and d~, and, therefore a unique v(s). Thispermits us to use v as the choice variable in period 1, which will make ourcalculations easier. Due to parties’ optimization,

∂πm

∂m= 0 and

∂πd

∂d= 0.

The uniqueness of m~ and d~ is then assured if:

∂2πm

[∂m]2≤ 0 and

∂2πd

[∂d]2≤ 0.

18

If we solve problems (A-7) and (A-8),

∂πm

∂m= (s−∆(R)− γ − δ)

αηR

m− ρ = 0 (A-9)

∂πd

∂d= (1− s+ ∆(R)− γ + δ)

α(1− η)Rd

− ω = 0. (A-10)

where we use (3) and (A-5) to derive that

∂R

∂m=αηR

mand

∂R

∂d=α(1− η)R

dand

d∆(R)dR

=2δR.

We can then use (A-9) and(A-10) to determine that m~ and d~ are uniqueas long as α ≤ 1/2. Specifically, if α ≤ 1/2,

∂2πm

[∂m]2=αηR

m2((s−∆(R)− γ − δ)(αη − 1)− 2δαη) ≤ 0

and

∂2πd

[∂d]2=α(1− η)R

d2((1− s+ ∆(R)− γ − δ)(α(1− η)− 1) + 2δ(2α(1− η)− 1)) ≤ 0,

where we infer from (A-9) that s −∆(R) − γ − δ ≥ 0 and from (A-8) that1− s+ ∆(R)− γ − δ ≥ 0.

Uniqueness of m~ and d~ then permits us to use v as an independentvariable, and find s = s(v), m(v), d(v) and R(v). If we combine (A-6), (A-9)and (A-10),

m

η=

(v − γ − δ)Rαρ

(A-11)

andd

1− η=

(1− v − γ + δ)Rαω

. (A-12)

We can then insert (A-11) and ((A-12)) into (3) to express R as a functionof v:

R(v) = λ

[αθ

(v − γ − δ

ρ

)η (1− v − γ + δ

ω

)1−η] α

1−α

. (A-13)

Combining (A-11), (A-12) and (A-13) then yields:

m(v) =(v − γ − δ)αη

ρR(v) (A-14)

19

andd(v) =

(1− v − γ + δ)α(1− η)ω

R(v). (A-15)

Period 1. In period 1, the multinational firm solves for the optimal contract(s, t). Since there is a unique v~ corresponding to s~, we can use v asthe choice variable. We can express the multinational firm’s maximizationproblem as:

maxv,t

πm = (v − γ + δ)R(v)− ρm(v) + t

subject to

πd = (1− v − γ − δ)R(v)− ωd(v)− t ≥ 0,

which simplifies to:

maxvπm = ((1− αη)(1− 2γ)− α(1− v − γ + δ))R(v) (A-16)

We can then solve (A-16) to obtain v~:

v~ = s− γ1− 2ψ2η − 1

+ δ (A-17)

where s is given by (10). From (A-6), we can obtain the corresponding s~

ass~ = v~ + ∆(R~)

Finally, we can insert (A-17) into (A-16) and combine with (20) to derivethe multinational firm’s profits:

π~m = π∗m = πm (1− 2γ)

11−α .

20

References

[1] Antras, P., 2003. Firms, contracts and trade structure. Quarterly Jour-nal of Economics 68, 1375–1418.

[2] Antras, P., 2005. Incomplete contracts and the product cycle. AmericanEconomic Review 45, 1054-1073.

[3] Antras, P., Helpman, E., 2004. Global sourcing. Journal of PoliticalEconomy 112, 552–580.

[4] Antras, P., Helpman, E., 2008. Contractual frictions and global sourc-ing, in: Helpman, E., Marin, D.,Verdier, T. (Eds.), The Organizationof Firms in a Global Economy, Harvard University Press, Cambridge.

[5] Bai, C., Tao, Z., Wu, C., 2004. Revenue sharing and control rightsin team production: theories and evidence from joint ventures. RandJournal of Economics 35, 277–305.

[6] Buckley, P., Casson, M., 1976. The Future of the Multinational Enter-prise. MacMillan, London.

[7] Caves, R., 1982. Multinational Enterprise and Economic Analysis.Cambridge University Press, New York.

[8] Desai, M., Moel, A., 2008. Czech mate: expropriation and investorprotection in a converging world. Review of Finance 12, 221–251.

[9] Dunning, J., 1977. Trade, location of economic activity and the MNE: asearch for an eclectic approach, in: Ohlin, B., Hesselborn, P., Wijkman,P. (Eds.), The International Allocation of Economic Activity, Holmes& Meyer, New York, pp. 395-418.

[10] Eden, L., Miller, S., 2004. Distance matters: liability of foreignness,institutional distance and ownership strategy, in: Hitt, M., Cheng J.(Eds.), The Evolving Theory of the Multinational Firm, Advances inInternational Management, Volume 16, Elsevier, Amsterdam, pp. 187-221.

[11] Grossman, G., Helpman, E., 2002. Integration versus outsourcing inindustry equilibrium. Quarterly Journal of Economics 117, 85–120.

[12] Grossman, G., Helpman, E., 2003. Outsourcing versus FDI in industryequilibrium. Journal of the European Economic Association 1, 317-327.

21

[13] Grossman, G., Helpman, E., 2005. Outsourcing in a global economy.Review of Economic Studies 72, 135-159.

[14] Helpman, E., 2006. Trade, FDI, and the organization of firms. Journalof Economic Literature 44, 589–630.

[15] Henisz, W., 2000. The institiutional environment for multinational in-vestment. The Journal of Law, Economics and Organization 16, 334–364.

[16] Hennart, J., 1982. A Theory of Multinational Enterprise. University ofMichigan Press, Ann Arbor.

[17] Hennart, J. 1988. A transaction cost theory of equity joint ventures.Strategic Management Journal 9, 361-374.

[18] Hennart, J., 1991. The transaction costs theory of joint ventures: anempirical study of japanese subsidiaries in the United States. Manage-ment Science 17, 483-497.

[19] Huang, Y., 2005. Are foreign firms privileged by their host govern-ments? Evidence from the 2000 world business environment survey.MIT Sloan Working Paper No. 4538-05.

[20] Hymer, S., 1976. The International Operations of National Firms: AStudy of Direct Investment. MIT Press, Cambridge.

[21] Inkpen, A., Beamish, P., 1997. Knowledge, bargaining power, and theinstability of international joint ventures. Academy of Management Re-view 22, 177-203.

[22] Luo, Y., 2002. Capability exploitation and building in a foreign market:implications for multinational enterprises. Organizational Studies 13,48-63.

[23] Melitz, M., 2003. The impact of trade on intra-industry reallocationsand aggregate industry productivity. Econometrica 71, 1695–1725.

[24] Perkins, S., Morck, R., Yeung, B., 2007. Innocents abroad: failure ratesof international joint ventures with pyramidal groups. Mimeo.

[25] Spencer, B., 2005. International outsourcing and incomplete contracts.Canadian Journal of Economics 38, 1107–1135.

22

[26] Wang, S.; Zhu, T., 2005. Control allocation, revenue sharing, and jointownership. International Economic Review 46, 895–915.

[27] Zaheer, S., 1995. Overcoming the liability of foreignness. The Academicof Management Journal 38, 341–363.

[28] Zaheer, S., 2002. The liability of foreignness, redux: a commentary.Journal of International Management 8, 351–358.

23

1/20

γ

1

1/2

γ−1

1

v

η

Perfect contractenforcement

Imperfect contractenforcement

Imperfect contractenforcement with

cronyism

δγ +

δγ +−1

Figure 1: Optimal ex post revenue distribution in an IJV

24

)1/( ααθ −

kmπ

Jmπ

Vmπ

)1/( ααθ − )1/(1

ααθ −

InternationalJoint Venture

Wholly-ownedsubsidiaryExit

)1/( ααθ −0

0

Jκ

Vκ

Figure 2: Productivity and Entry Mode

25